library(DiagrammeR)
gp_nodes = create_nodes(
nodes = c('GP w/ Matern', 'GP w/ Matern & ν fixed', 'GP w/ Matern & ν = ∞','GP w/ SqExp',
'RKHS', 'AR(1)', 'OU'),
type = 'a',
value = 1,
label = TRUE,
style = 'filled',
color = 'papayawhip',
shape = 'circle',
fixedsize=T,
distortion='',
fillcolor='',
fixedsize='',
fontcolor='',
fontname='',
fontsize=3,
height='',
penwidth='',
peripheries='',
shape='',
sides='',
# tooltip='FU',
width='',
x='',
y='')
edges = create_edges(
from = '',
to = '',
rel = '',
distortion='',
fillcolor='',
fixedsize='',
fontcolor='',
fontname='',
fontsize='',
height='',
penwidth='',
peripheries='',
shape='',
sides='',
style='',
tooltip='',
width='',
x='',
y=''
)
render_graph(create_graph(gp_nodes))
GP = Gaussian process SqExp = squared exponential covariance structure Matern = Matern covariances structure OU = Ornstein-Uhlenbeck \(\mathcal{l}\) = horizontal length-scale \(\mathcal{\nu}\) = controls differentiability
Rasmussen & Williams (2006). Gaussian Processes for Machine Learning. Murphy (2012). Machine Learning: A probabilistic perspective.